Algorithms Aren't Working for Me — An Anti-Algorithm Manifesto

Algorithms Don't Work for Me

At some point I started noticing something off. These algorithms are supposed to be tailored to me, but the more I used them, the narrower my life felt. Things got more convenient, sure — but that convenience kept nudging me somewhere I hadn't chosen to go. So I sat down and worked through a few of these experiences. The short version: algorithms don't work for me. They work for a company's metrics. And that metric is almost always "how long can I keep this person on the app."

Dating Apps: Optimized for Return Visits, Not Matches

Use a dating app for a while and you'll notice a strange pattern. It never quite gives you someone who's actually a great match. Instead, it keeps surfacing people who are close but slightly off — good enough to keep you curious, not good enough to close the loop. I used to think my preference settings were wrong. Thinking about it more, though, it's the obvious outcome. If the app matched me with someone genuinely great right away, I'd hit it off with them and delete the app. From the platform's side, that's the worst-case scenario. Keep feeding me near-misses instead, and I keep telling myself "maybe this time," and keep opening the app. What a dating app's algorithm actually optimizes for isn't "a good match" — it's "return rate." The moment I actually meet someone great and delete the app, that app loses revenue. Not solving my problem is the business model.

Social Media: Too Much Information, No Way to Digest It

I wanted to learn more about marketing, so I made a new social media account. A few days of reading marketing and AI-related posts, following a handful of accounts, and the entire feed reorganized itself around exactly that. Here's the problem: the volume of marketing- and AI-related posts pouring in every single day was more than I could possibly read. Semiconductors have advanced at an insane pace over the last 50 years. But the device I use to take in information is still a keyboard, a mouse, and my own eyes. Processing power scaled exponentially; the bandwidth of the human input/output device did not. The algorithm doesn't care about that mismatch. If I've shown interest in a topic, it keeps pushing more of it at me regardless of whether I can actually process it. The result is that I'm not really "consuming" information anymore — I'm being buried under it.

Netflix After a Breakup: Turning Sadness Up, Not Down

Someone I know went through a breakup not long ago. Open YouTube or Netflix, and — strangely — only sad romance movies and breakup-themed videos kept surfacing. It's the algorithm reading recently watched content, search terms, and time spent, then recommending "whatever this person is most likely to get absorbed in right now." It's true that when you're sad, sad content pulls you in more. The problem is that pull isn't actually good for you. Going through a breakup doesn't mean you should only watch sad movies. Sometimes you need a dumb comedy, or a horror movie, just to reset your mood. But the algorithm keeps serving up "whatever has the highest engagement right now" — and that's rarely the thing that actually helps. It doesn't move you toward recovery; it just keeps you locked in that emotional state a little longer. From a watch-time perspective, a sad person watching sad content for hours is a great outcome.

One Hip-Hop Song, and Suddenly It's Hip-Hop All Day

Music streaming works the same way. Play one hip-hop track by accident, and the entire recommendation feed flips to hip-hop for the rest of the day. Some days I want rock. Some days I want jazz. But the algorithm decides "this person likes hip-hop now" and keeps pushing exactly that. Taste is supposed to wander in different directions — the algorithm treats that wandering as noise and tries to collapse it onto a single axis. Do that enough times, and your taste actually does narrow. What used to be varied ends up hardening into whatever the algorithm kept showing you.

The Old Newspaper Forced Some Breadth on You

Back when people read newspapers or watched the evening news, articles you had zero interest in still crossed your eyes whether you liked it or not. You'd glance at a front-page political headline even without caring about politics; you'd catch an exchange-rate mention even without caring about economics. It was a structure that forced a baseline of general knowledge on you, whether you wanted it or not. That's not how it works anymore. Mobile algorithms only ever show you what you're already interested in. Anything outside that never even shows up in the feed. It looks convenient, but it quietly removes the chance to pick up the kind of "knowledge you didn't ask for" that used to come naturally. Here's how the two structures compare.

	Then (newspapers / TV news)	Now (algorithmic feed)
Who decides what you see	Editors, journalists	Your own past behavioral data
Scope of exposure	Forces exposure to topics outside your interest	Repeats only what you're already interested in
Goal	Inform you as a citizen	Maximize time-on-platform and clicks
Result	Broad, shallow general knowledge	Narrow, deep bias

The Harm Algorithms Cause, Summed Up

There's one pattern running through all of these.

Dating apps — optimize for return visits, not good matches
Social media overload — keep pushing more information than a human can process, just because it matches your interests
Emotion-based recommendations — amplify your mood by serving more sad content when you're sad, more angry content when you're angry
Narrowing taste — treat one choice as your entire preference, and strip out diversity as noise

All four wear the label "personalization" on the surface. What's actually being optimized underneath isn't my happiness or my growth — it's time-on-platform, return rate, click-through rate. From the algorithm's perspective, a diverse, balanced life is actually a loss. Diversity makes you harder to predict; harder to predict means lower recommendation accuracy; lower accuracy means less time spent on the app.

Humans Don't Run in Parallel

The real problem underneath all of this is time. Computers add more cores and run more tasks simultaneously. People don't work that way. A day has 24 hours, and the slice of that you can actually spend focused, learning, and feeling things is much smaller. Yet the algorithm keeps signaling "you need to learn this too," "you'll fall behind if you don't." The same way one new marketing account buried me in marketing and AI content all day, algorithms try to pour my remaining time entirely into the one topic I've already shown interest in. But a person needs that time for other things too — learning something unrelated, feeling a different emotion, meeting a different person. The algorithm doesn't factor any of that in. There's no reason for it to. None of that shows up in the metrics of the company that built it.

Which Is Why I Built a Tape App and ChainPlay

This exact concern is part of why I built the Cassette Music Player and ChainPlay. The cassette app wasn't about algorithm-recommended tracks — I wanted to recreate the experience of loading only the songs I chose myself onto a tape, in order, and listening straight through without skipping. ChainPlay comes from the same place. Open YouTube and autoplay plus recommendations never stop — I just wanted to watch a handful of videos I'd picked myself, in the order I picked them. So I built an app that lets you string YouTube links into a "chain," where only the videos you added play, in that order, and then it stops. Neither app does much, feature-wise. But the direction is clear — not being pulled along by whatever the algorithm recommends, but listening to and watching what a person actually chose and made. I think that direction only gets more important as AI gets smarter. The stronger AI's grip on deciding what gets recommended, the more deliberately I need to hold onto deciding what I actually watch.

So, Anti-Algorithm

The conclusion here isn't "cut out algorithms entirely." That's not realistic. But I think it's worth at least knowing that what an algorithm shows you isn't "the best thing for you" — it's "the optimal thing for the company." If a dating app keeps showing you near-misses, question it once in a while. If your feed has narrowed to one topic, go search for something else on purpose. When you're in a bad mood, watch something you picked yourself instead of whatever the algorithm hands you. If it's hip-hop on repeat, go search out some rock every once in a while. Algorithms are smart. But that intelligence isn't pointed at me. Just knowing that much is enough to use them a little differently.